Biases in Israel

Self-Reported Direct Preference

Participants reported their preference between the dominant and non-dominant category (scale: 1 = “I strongly prefer NON-DOMINANT CATEGORY to DOMINANT CATEGORY to 7 =”I strongly prefer DOMINANT CATEGORY to NON-DOMINANT CATEGORY").

Note: The plot below shows the effect-size (Cohen’s d) as a function of the direct self-reported preference task. Larger effect-sizes indicate a stronger pro-dominant bias. Not all tasks included the direct self-reported preference measure.

aget <- ttestOneS(age, vars = vars(att), testValue = 4, effectSize = T)
aget <- as.data.frame(aget$ttest)
aged <- round(aget[[6]], 2)

racet <- ttestOneS(race, vars = vars(att), testValue = 4, effectSize = T)
racet <- as.data.frame(racet$ttest)
raced <- round(racet[[6]], 2)

weightt <- ttestOneS(weight, vars = vars(att), testValue = 4, 
    effectSize = T)
weightt <- as.data.frame(weightt$ttest)
weightd <- round(weightt[[6]], 2)

ethnict <- ttestOneS(ethnic, vars = vars(att), testValue = 4, 
    effectSize = T)
ethnict <- as.data.frame(ethnict$ttest)
ethnicd <- round(ethnict[[6]], 2)

sexualityt <- ttestOneS(sexuality, vars = vars(att), testValue = 4, 
    effectSize = T)
sexualityt <- as.data.frame(sexualityt$ttest)
sexualityd <- round(sexualityt[[6]], 2)

usat <- ttestOneS(usa, vars = vars(att), testValue = 4, effectSize = T)
usat <- as.data.frame(usat$ttest)
usad <- round(usat[[6]], 2)


all_ds <- c(aged, raced, weightd, ethnicd, sexualityd, usad) %>% 
    as.data.frame()
colnames(all_ds) <- "es"

all_ds$Task <- c("Age", "Race", "Weight", "Ethnicity", "Sexuality", 
    "Nationalism")

pp <- ggdotchart(all_ds, x = "Task", y = "es",
           color = "Task",                                # Color by groups
           ylab = "Cohen's d",
           #palette = c("#00AFBB", "#E7B800", "#FC4E07"), # Custom color palette
           sorting = "descending",                       # Sort value in descending order
           add = "segments",                             # Add segments from y = 0 to dots
           rotate = TRUE,                                # Rotate vertically
           #group = "cyl",                                # Order by groups
           dot.size = 30,                                 # Large dot size
           label = round(all_ds$es,3),                        # Add mpg values as dot labels
           font.label = list(color = "white", size = 25, 
                             vjust = 0.5),               # Adjust label parameters
           ggtheme = theme_pubr()                        # ggplot2 theme
           )

pp <- pp + geom_hline(yintercept = 0, linetype = "dashed", color = "darkblue", 
    size = 1.2)

pp <- pp + font("title", size = 30, color = "black", face = "bold")+
font("legend.text", size = 26, color = "black", face = "bold")+
font("xlab", size = 28, color = "black", face = "bold")+
font("ylab", size = 28, color = "black", face = "bold")+
font("xy.text", size = 25, color = "black", face = "bold")
pp <- pp +theme(legend.position="none")

pp

Feeling thermometer (Difference score)

Participants reported their feelings toward each category (scale: 0 “very cold” to 10 “very warm”). We computed a difference score of [dominant category - non-dominant category] as another measure of self-reported preference.
Gender-Science: In this study the scale was 1 = “Strongly female” to 7 = “Strongly male”. The score was computed as the rating for science minus the rating for liberal arts. Higher scores indicate a consideration of science as masculine and liberal arts as feminine.

The effect-sizes on this measure were smaller, compared with the previous measure. This difference might result from the nature of each measure: while the Self-reported preference measure directly compares the two categories (as does the IAT), the thermometer does not compare the two categories directly. A comparative context might facilitate bias.

Note: The plot below shows the effect-size (Cohen’s d) as a function of the feeling thermometer task. Larger effect-sizes indicate a stronger pro-dominant bias. Not all tasks included the feeling thermometer measure.

aget <- ttestOneS(age, vars = vars(diff_thermo), testValue = 0, 
    effectSize = T)
aget <- as.data.frame(aget$ttest)
aged <- round(aget[[6]], 2)

racet <- ttestOneS(race, vars = vars(diff_thermo), testValue = 0, 
    effectSize = T)
racet <- as.data.frame(racet$ttest)
raced <- round(racet[[6]], 2)

skint <- ttestOneS(skin, vars = vars(diff_thermo), testValue = 0, 
    effectSize = T)
skint <- as.data.frame(skint$ttest)
skind <- round(skint[[6]], 2)

weightt <- ttestOneS(weight, vars = vars(diff_thermo), testValue = 0, 
    effectSize = T)
weightt <- as.data.frame(weightt$ttest)
weightd <- round(weightt[[6]], 2)

usat <- ttestOneS(usa, vars = vars(diff_thermo), testValue = 0, 
    effectSize = T)
usat <- as.data.frame(usat$ttest)
usad <- round(usat[[6]], 2)

gender_sciencet <- ttestOneS(gender_science, vars = vars(diff_thermo), 
    testValue = 0, effectSize = T)
gender_sciencet <- as.data.frame(gender_sciencet$ttest)
gender_scienced <- round(gender_sciencet[[6]], 2)

all_ds <- c(aged, raced, skind, weightd, usad, gender_scienced) %>% 
    as.data.frame()
colnames(all_ds) <- "es"

all_ds$Task <- c("Age", "Race", "Skin", "Weight", "Nationalism", 
    "Gender-Science")

pp <- ggdotchart(all_ds, x = "Task", y = "es",
           color = "Task",                                # Color by groups
           ylab = "Cohen's d",
           #palette = c("#00AFBB", "#E7B800", "#FC4E07"), # Custom color palette
           sorting = "descending",                       # Sort value in descending order
           add = "segments",                             # Add segments from y = 0 to dots
           rotate = TRUE,                                # Rotate vertically
           #group = "cyl",                                # Order by groups
           dot.size = 30,                                 # Large dot size
           label = round(all_ds$es,3),                        # Add mpg values as dot labels
           font.label = list(color = "white", size = 25, 
                             vjust = 0.5),               # Adjust label parameters
           ggtheme = theme_pubr()                        # ggplot2 theme
           )

pp <- pp + geom_hline(yintercept = 0, linetype = "dashed", color = "darkblue", 
    size = 1.2)

pp <- pp + font("title", size = 30, color = "black", face = "bold")+
font("legend.text", size = 26, color = "black", face = "bold")+
font("xlab", size = 28, color = "black", face = "bold")+
font("ylab", size = 28, color = "black", face = "bold")+
font("xy.text", size = 25, color = "black", face = "bold")
pp <- pp +theme(legend.position="none")

pp

Does that mean that Israelis are more OK with being biased toward Fat people, and in favor of their own country, than toward elderly people, for example? Does that relate to certain social values in the Israeli society? We think these are interesting questions that could be addressed in future research.

Participation per year and task:

# make N per year & task table

yy <- all_iat %>% group_by(Year, Task) %>% tally() %>% pivot_wider(names_from = Task, 
    values_from = c(n))
yy %>% kable() %>% kable_styling(bootstrap_options = c("striped", 
    "hover", "condensed", full_width = F, position = "left"))

# iat

get_es_iat <- function(x) {
    es <- x$IAT %>% cohens_d()
    es <- round((es), 2)
    es_list <- list(es = es)
}

df <- all_iat %>% drop_na() %>% group_by(Task, Year) %>% nest() %>% 
    drop_na()

iat_df <- df %>% mutate(params = purrr::map(.x = data, .f = ~get_es_iat(.x))) %>% 
    unnest_wider(params) %>% select(-data) %>% unnest_wider(es)

# self-reported preference

get_es_sr <- function(x) {
    es <- x$att_rs %>% cohens_d()
    es <- round((es), 2)
    es_list <- list(es = es)
}

df <- all_att %>% drop_na() %>% group_by(Task, Year) %>% nest() %>% 
    drop_na()

att_df <- df %>% mutate(params = purrr::map(.x = data, .f = ~get_es_sr(.x))) %>% 
    unnest_wider(params) %>% select(-data) %>% unnest_wider(es)

# self-reported thermometer

get_es_thr <- function(x) {
    es <- x$diff_thermo %>% cohens_d()
    es <- round((es), 2)
    es_list <- list(es = es)
}

df <- all_diff_thermo %>% drop_na() %>% group_by(Task, Year) %>% 
    nest() %>% drop_na()

thermo_df <- df %>% mutate(params = purrr::map(.x = data, .f = ~get_es_thr(.x))) %>% 
    unnest_wider(params) %>% select(-data) %>% unnest_wider(es)

Note: The plot below shows the effect-size (Cohen’s d) as a function of year and task. Larger effect-sizes indicate a stronger pro-dominant bias.

IAT

iat_df$Year = factor(iat_df$Year, c("2009", "2010", "2011", "2012", 
    "2013", "2014", "2015", "2016", "2017", "2018", "2019"))

pp <- ggline(iat_df, x = "Year", y = "Cohens_d", ylab = "Cohen's d", 
    color = "Task", size = 2)

pp <- pp + geom_hline(yintercept = 0, linetype = "dashed", color = "darkblue", 
    size = 1.2)

pp <- pp + font("title", size = 20, color = "black", face = "bold") + 
    font("legend.text", size = 25, color = "black", face = "bold") + 
    font("xlab", size = 28, color = "black", face = "bold") + 
    font("ylab", size = 28, color = "black", face = "bold") + 
    font("xy.text", size = 25, color = "black", face = "bold")
# pp <- pp +theme(legend.position='none')

pp

Self-Reported Direct Preference

att_df$Year = factor(att_df$Year, c("2009", "2010", "2011", "2012", 
    "2013", "2014", "2015", "2016", "2017", "2018", "2019"))

pp <- ggline(att_df, x = "Year", y = "Cohens_d", ylab = "Cohen's d", 
    color = "Task", size = 2)

pp <- pp + geom_hline(yintercept = 0, linetype = "dashed", color = "darkblue", 
    size = 1.2)

pp <- pp + font("title", size = 20, color = "black", face = "bold") + 
    font("legend.text", size = 25, color = "black", face = "bold") + 
    font("xlab", size = 28, color = "black", face = "bold") + 
    font("ylab", size = 28, color = "black", face = "bold") + 
    font("xy.text", size = 25, color = "black", face = "bold")
# pp <- pp +theme(legend.position='none')

pp

Self-Report: Feeling Thermometer Difference

thermo_df$Year = factor(thermo_df$Year, c("2009", "2010", "2011", 
    "2012", "2013", "2014", "2015", "2016", "2017", "2018", "2019"))

pp <- ggline(thermo_df, x = "Year", y = "Cohens_d", ylab = "Cohen's d", 
    color = "Task", size = 2)

pp <- pp + geom_hline(yintercept = 0, linetype = "dashed", color = "darkblue", 
    size = 1.2)

pp <- pp + font("title", size = 20, color = "black", face = "bold") + 
    font("legend.text", size = 25, color = "black", face = "bold") + 
    font("xlab", size = 28, color = "black", face = "bold") + 
    font("ylab", size = 28, color = "black", face = "bold") + 
    font("xy.text", size = 25, color = "black", face = "bold")
# pp <- pp +theme(legend.position='none')

pp